Wave-Based Dispatch for Circuit Cutting in Hybrid HPC--Quantum Systems

A team from CESGA (Ricard S. García-Raigada, Josep Jorba, Sergio Iserte) has published a paper on arXiv introducing DQR (Dynamic Queue Router), a novel runtime framework designed to solve a critical bottleneck in hybrid High-Performance Computing (HPC)-quantum systems. Current methods for running large quantum algorithms on today's Noisy Intermediate-Scale Quantum (NISQ) hardware often rely on 'circuit cutting,' which decomposes a large circuit into smaller, executable fragments. However, existing frameworks tightly couple the logic for cutting circuits with the orchestration of their execution, preventing HPC centers from applying their mature, high-efficiency resource management policies to quantum workloads. DQR bridges this gap by treating these quantum circuit fragments as standard, schedulable computational tasks.

DQR's architecture is built on three key innovations: a backend-agnostic fragment descriptor that exposes a circuit's structural properties without requiring execution layers to parse complex quantum code; a 'wave-based' coordinator that uses non-blocking polling to achieve pipeline concurrency; and a production-ready implementation on the CESGA Qmio supercomputer. This implementation seamlessly integrates both local on-premises quantum processing units (QPUs) and remote cloud backends like IBM's Torino. In experiments running a 32-qubit Hardware-Efficient Ansatz circuit, the framework not only improved overall makespan but also demonstrated robust, transparent failover recovery. It could automatically reroute tasks from an incompatible local QPU to classical simulators without needing to restart the entire computational pipeline. For deeper circuits, the coordination overhead—the 'cost' of managing this hybrid system—was measured at just 5% of total execution time, proving the solution's scalability.